Intel announced on Wednesday the creation of Hala Point, its latest neuromorphic computer system. Using 1,152 Loihi 2 processors, it's designed to aid in researching future brain-inspired artificial intelligence and developing more sustainable uses of today's AI.

Though introduced today, Hala Point will not be commercially available — it's a research prototype. Intel has deployed the neuromorphic system to Sandia National Laboratories, part of the U.S. Department of Energy's National Nuclear Security Administration (NNSA). This deployment is part of a relationship between the two parties that goes back to 2021 to explore neuromorphic computing in AI further.

"The computing cost of today's AI models is rising at unsustainable rates. The industry needs fundamentally new approaches capable of scaling," Mike Davies, Intel Labs' director of its Neuromorphic Computing Lab, said in a statement. "For that reason, we developed Hala Point, which combines deep learning efficiency with novel brain-inspired learning and optimization capabilities."

Intel is characterizing Hala Point as supporting up to 30 quadrillion operations per second or 30 petaops, with an efficiency exceeding 15 trillion 8-bit operations per second per watt "when executing conventional deep neural networks."

Along with its thousands of Loihi 2 processors, Hala Point supports up to 1.15 billion neurons and 128 billion synapses distributed over more than 140,000 neuromorphic processing cores. It also includes more than 2,300 embedded x86 processors. It can also provide 16 petabytes per second of memory bandwidth, 11 PB/s of inter-core communication bandwidth and 5.5 terabytes per second of inter-chip communication bandwidth.

The architecture of Intel's Hala Point neuromorphic computing system. Credit: Intel

It's an evolution from the company's first large-scale research system, Pohoiki Springs, in that it has greater neuron capacity (10x) and higher performance (12x).

"Applied to bio-inspired spiking neural network models, Hala Point can execute its full capacity of 1.15 billion neurons 20 times faster than a human brain and up to 200 times faster rates at lower capacity," Davies remarks. "While Hala Point is not intended for neuroscience modeling, its neuron capacity is roughly equivalent to that of an owl brain or the cortex of a capuchin monkey."

The work done to advance Pohoiki Springs, combined with the architectural improvements made to Loihi 2, allows Hala Point to "bring neuromorphic performance and efficiency gains to mainstream conventional deep learning models" such as those processing real-time workloads (video, speech and wireless communications).

Since it's not available to the public, what does this neuromorphic computing system do? Sandia Labs and the NNSA research teams are believed to be using it to "realize brain-based computing on a large scale." It may also eventually tackle large-scale problems in physics, chemistry and the environment.

Intel's Hala Point. Credit: Intel

"Hala Point can perform solve optimization problems using 100 times less energy at speeds as much as 50 times faster than conventional CPU and GPU architectures," Davies explains. "This is an exciting new domain of research where the algorithms running on Loihi 2 are highly brain-inspired, looking dramatically different from the leading algorithms designed for CPU and GPU architectures. Applications of such optimization capabilities might include logistics, vehicle fleet routing, railway scheduling, smart city infrastructure management and other similar scheduling/planning/search workloads."

Intel says it's unable to disclose how much Hala Point costs but reveals access is given to "a variety of smaller-scale systems to members of the Intel Neuromorphic Research Community via its cloud platform." That program is free and open to those in academic, government and corporate roles.